ERIC ED610960: Advancing Minority Gifted Identification: Evidence from a Randomized Trial of Nurturing for a Bright Tomorrow

2017·0.04 MB·English

by  ERIC

SREE Spring 2017 Conference

Title:
“Advancing Minority Gifted Identification: Evidence from a Randomized Trial of Nurturing for
a Bright Tomorrow”

Authors:
Angel L. Harris, Duke University
Darryl V. Hill, Wake County Public School System
Matthew A. Lenard, Wake County Public School System

Abstract
This paper reports preliminary results from a randomized controlled trial (RCT) of a multi-
component nurturing program designed to increase gifted identification among minority
students. In 2014-15, the Wake County Public School System (WCPSS), in partnership with
Duke University, implemented Nurturing for a Bright Tomorrow (“Nurturing”) in response to
chronically-low gifted identification among Black and Hispanic students in two-thirds of the
district’s elementary schools. This under-representation mirrored national and state trends.
According to the U.S. Department of Education’s Office of Civil Rights, while 40% of Black and
Hispanic students recently were enrolled in schools offering gifted education programs, only
28% ultimately enrolled (Lhamon, 2015). In more than 40 states, Black and Hispanic students
were underrepresented in gifted and talented programs (Yoon & Gentry, 2009). And in various
studies at the school, district, and state levels, racial and ethnic gaps in representation are
reported at various magnitudes (Carman & Taylor, 2009; Lewis, DeCamp-Fritson, Ramage,
McFarland, & Archwamety, 2007; McBee, 2010; Naglieri & Ford, 2003; Olszewski-Kubilius &
Lee, 2011).

The purpose of Nurturing is to train K-2 teachers to develop the skills and expectations required
to help children attain gifted status. A large body of evidence supports the belief that variations
in teacher disposition toward students can significantly influence perceived or real student
outcomes (Anderson-Clark, Green, & Henley, 2008; Dee, 2004, 2005; Grissom & Redding,
2016; Love & Kruger, 2005; McKown & Weinstein, 2008). Nurturing is designed to influence
these dispositions through a comprehensive curricular approach that includes three components:
Thinking Skills & Key Concepts (Parks & Black, 1997), Habits of Mind (Costa & Kallick,
2005), and Task Rotations (Silver, Jackson, & Moirao, 2011). Taken together, these three
approaches provide teachers with a framework to differentiate instruction, teach advanced
vocabulary and speaking skills, and build sustainable approaches to problem solving.

Two previous iterations of Nurturing were associated with improved achievement and gifted
identification, but neither produced causal estimates (Watson & Darity, 2010). In this current
iteration, we randomly assigned Nurturing to 16 elementary schools on the basis of gifted
identification rates in math and reading. We first identified all elementary schools below the
district rate of 4% and recruited 32 schools not previously exposed to Nurturing to join the full
analytic sample. We then sorted schools on these rates and randomly assigned the program
within pairs. Table 1 shows balance between the treatment and control groups on a variety of
pre-treatment characteristics. The business-as-usual condition at the 16 control schools includes
1
U-STARS~PLUS (Coleman & Job, 2014) and Primary Education Thinking Skills (PETS)
(Nichols, 1997).

Our main outcome of interest is the Naglieri Nonverbal Ability Test (NNAT2), administered to
1st graders in spring 2016 (who began Nurturing as kindergarten students in fall 2015). We use
NNAT2 as an interim measure of gifted potential prior to formal identification in grade 3 (fall
2017). Studies of the NNAT suggest that the test strongly predicted gifted identification for
diverse populations (Lidz & Macrine, 2001; Naglieri & Ford, 2003) and performed similarly to
the CogAT6 achievement test (Giessman, Gambrell, & Stebbins, 2013).

Our sample consists of roughly 3,500 1st grade students who took NNAT2 in spring 2016. We
examine the impact of Nurturing on the NNAT2 in three ways: (1) on the NNAT2 scale score
(Tables 2-3), (2) on the odds of being classified as gifted by scoring two standard deviations
(130) above the NNAT2 mean (100) (Table 4), and (3) on the school level counts of gifted
students (Table 5). For (1), we use a two-level random effects model controlling for prior
achievement and a various student-level characteristics. For (2), we use logistic regression to
estimate the odds of being identified. Finally, for (3), we use Poisson and negative binomial
regression models at the school-level (N = 32) to measure gifted counts at our unit of assignment.
Due to the small sample size in our third set of outcomes, we fit separate models to account for
extreme value influence on the basis of Cook’s D statistic thresholds and by Winsorizing the
data.

Results suggest that Nurturing had small-to-moderate positive impacts on NNAT2 scores,
student-level identification, and school-level identification. Across a range of model
specifications in Table 2, Nurturing impacts on NNAT2 scores ranged from 0.05 to 0.10.
Table 3 shows that neither Black nor Hispanic students outperformed their control group
counterparts on NNAT2. Table 4 shows that the odds of being identified as gifted are higher for
the treatment group than for the control group [1.9, 2.2]. Finally, we examined whether
Nurturing increased the number of gifted students at the school-level (Table 5). The full Model
(1) shows that treatment group schools had an incidence rate for counts that was expected to
change by a factor of 4.1 (p < .01). This effect is potentially influenced by a single school at
which 17 students met the 130 cutoff. Removing this and a second school on the basis of forming
an outlying cluster (D = [0.89, 0.95]) resulted in similar rate ratios. The treatment group in the
Winsorized sample had a rate ratio for counts expected to change by a factor of 2.3 (p < .01).

WCPSS’s RCT of Nurturing with Duke University comes at a time when school districts
nationwide are under increased scrutiny for low gifted identification rates among minority
students. These interim results suggest that Nurturing is having a modest positive impact on key
test outcome of gifted identification and large impacts on school-based counts. Despite our
cautious optimism that Nurturing holds promise for all students attending schools that
historically under-identify, the program does not yet appear to provide marginal benefits to Black
and Hispanic students. These interim results, however, suggest that implementation is on the
right track. We will present final results during the 2017-18 school year when the first cohort
formally tests for gifted status.

2
Appendix – References and Tables

References

Anderson-Clark, T. N., Green, R. J. & Henley, T. B. (2008). The relationship between first
names and teacher expectations for achievement motivation. Journal of Language and Social
Psychology, 27(1), 94–99.

Carman, C. A. & Taylor, D. K. (2009). Socioeconomic status effects on using the Naglieri
Nonverbal Ability Test (NNAT) to identify the gifted/talented. Gifted Child Quarterly.

Coleman, M. R. & Job, J. (2014). U-STARS~ PLUS science & nonfiction connections. Council
For Exceptional Children.

Costa, A. L. & Kallick, B. (2005). Habits of mind.

Dee, T. S. (2004). Teachers, Race, and Student Achievement in a Randomized Experiment.
Review of Economics and Statistics, 86(1), 195–210.

Dee, T. S. (2005). A teacher like me: Does race, ethnicity, or gender matter? The American
Economic Review, 95(2), 158–165.

Giessman, J. A., Gambrell, J. L. & Stebbins, M. S. (2013). Minority Performance on the Naglieri
Nonverbal Ability Test, Versus the Cognitive Abilities Test, Form 6 One Gifted Program’s
Experience. Gifted Child Quarterly, 57(2), 101–109.

Grissom, J. A. & Redding, C. (2016). Discretion and Disproportionality. AERA Open, 2(1),
2332858415622175.

Lewis, J. D., DeCamp-Fritson, S. S., Ramage, J. C., McFarland, M. A. & Archwamety, T.
(2007). Selecting for ethnically diverse children who may be gifted using Raven’s standard
progressive matrices and Naglieri nonverbal abilities test. Multicultural Education, 15(1), 38.

Lhamon, C. E. (2015). Protecting Civil Rights, Advancing Equity. U.S. Department of Education
Office of Civil Rights.

Lidz, C. S. & Macrine, S. L. (2001). An Alternative Approach to the Identification of Gifted
Culturally and Linguistically Diverse Learners The Contribution of Dynamic Assessment.
School Psychology International, 22(1), 74–96.

Love, A. & Kruger, A. C. (2005). Teacher beliefs and student achievement in urban schools
serving African American students. The Journal of Educational Research, 99(2), 87–98.

McBee, M. (2010). Examining the probability of identification for gifted programs for students
in Georgia elementary schools: A multilevel path analysis study. Gifted Child Quarterly,
3
54(4), 283–297.

McKown, C. & Weinstein, R. S. (2008). Teacher expectations, classroom context, and the
achievement gap. Journal of School Psychology, 46(3), 235–61. doi:10.1016/j.jsp.2007.05.001

Naglieri, J. A. & Ford, D. Y. (2003). Addressing underrepresentation of gifted minority children
using the Naglieri Nonverbal Ability Test (NNAT). Gifted Child Quarterly, 47(2), 155–160.

Nichols, J. (1997). Primary Education Thinking Skills\# 3.

Olszewski-Kubilius, P. & Lee, S.-Y. (2011). Gender and other group differences in performance
on off-level tests: Changes in the 21st century. Gifted Child Quarterly, 55(1), 54–73.

Parks, S. & Black, H. (1997). Building thinking skills. Critical Thinking Books \& Software.

Silver, H. F., Jackson, J. W. & Moirao, D. R. (2011). Task rotation: Strategies for differentiating
activities and assessments by learning style. ASCD.

Watson, M. & Darity, W. (2010). Project Bright Idea 2: Interest Development Early Abilities.

Yoon, S. Y. & Gentry, M. (2009). Racial and ethnic representation in gifted programs: Current
status of and implications for gifted Asian American students. Gifted Child Quarterly.

4
Table 1. Pre-Intervention Balance between Treatment and Control Groups
Treatment Control Difference p-
Variable Group Group (T - C) value
SWD 0.068 0.064 0.004 0.714
Male 0.526 0.513 0.013 0.370
LEP 0.135 0.115 0.020 0.444
American Indian/AK Native 0.005 0.003 0.002 0.283
Hispanic 0.270 0.216 0.054 0.122
Black 0.326 0.340 -0.014 0.822
Multiracial 0.028 0.032 -0.004 0.451
Native HI/Pacific Islander 0.003 0.001 0.002 0.136
Asian 0.039 0.042 -0.003 0.881
SES 0.087 0.072 0.015 0.201
Prior FSF Score 11.2 11.4 -0.226 0.828
Prior LNF Score 19.0 20.2 -1.186 0.372
N (schools) 16 16
n (students) 2367 2326
Note:
T-C: Treatment group mean minus control group mean.
FSF: DIBELS First Sound Fluency; LNF: DIBELS Letter Naming Fluency
Student-level means calculated using mixed-effects regression with robust standard errors.
* p<0.05 ** p<0.01 *** p<0.001

5
Table 2. Impact of Nurturing for a Bright Tomorrow on NNAT2 Scale Scores
(1) (2) (3) (4) (5)
Nurturing 0.050 0.068 0.100** 0.088* 0.064
(0.101) (0.049) (0.042) (0.050) (0.047)
SWD -0.366*** -0.366*** -0.358*** -0.335***
(0.059) (0.059) (0.064) (0.067)
Male 0.027 0.027 0.065** 0.070**
(0.031) (0.031) (0.032) (0.034)
LEP -0.315*** -0.310*** -0.142** -0.140**
(0.053) (0.053) (0.057) (0.062)
Am-Ind/AK -0.200 -0.207 -0.084 -0.088
(0.253) (0.253) (0.274) (0.271)
Hispanic -0.098* -0.087 0.018 -0.009
(0.052) (0.053) (0.055) (0.059)
Black -0.670*** -0.645*** -0.634*** -0.658***
(0.044) (0.045) (0.047) (0.050)
Multiracial -0.226** -0.219** -0.223** -0.265**
(0.095) (0.096) (0.105) (0.110)
HI/PI 0.160 0.152 0.969 0.855
(0.521) (0.521) (0.855) (0.847)
Asian 0.163** 0.170** 0.206** 0.166*
(0.080) (0.082) (0.090) (0.094)
SES -0.322*** -0.309*** -0.199*** -0.170***
(0.038) (0.038) (0.040) (0.043)
FSF 0.014*** 0.013***
(0.002) (0.002)
LNF 0.011*** 0.010***
(0.001) (0.001)
Constant -0.042 0.453*** 1.345** 1.370* 1.248*
(0.072) (0.046) (0.646) (0.747) (0.707)
School-Level Cov N N Y Y Y
KIA N N N N Y
Constant () 0.268*** 0.107*** 0.057** 0.083*** 0.065**
(0.038) (0.023) (0.026) (0.025) (0.029)
Constant () 0.963*** 0.899*** 0.899*** 0.851*** 0.840***
(0.012) (0.011) (0.011) (0.011) (0.012)
Observations 3435 3435 3435 2868 2537
Notes: “LNF” is DIBELS Letter Naming Fluency. “FSF” is DIBELS First Sound Fluency. “KIA” is the
Kindergarten Initial Assessment. “School-Level Cov” represents student-level covariates aggregated up to the
school level. Standard errors in parentheses
* p < .10, ** p < .05, *** p < .01

6
Table 3. Subgroup Impacts of Nurturing for a Bright Tomorrow on NNAT2 Scale Scores
(1) (2) (3) (4)
Black Hispanic Black Male Hispanic Male
Nurturing 0.011 0.111 -0.051 0.110
(0.068) (0.073) (0.092) (0.143)
SWD -0.366*** -0.367*** -0.366*** -0.365***
(0.059) (0.059) (0.059) (0.059)
Male 0.027 0.028 0.000 0.000
(0.031) (0.031) (.) (.)
LEP -0.310*** -0.306*** -0.314*** -0.313***
(0.053) (0.053) (0.053) (0.053)
Am-Ind/AK -0.207 -0.204 -0.202 -0.201
(0.253) (0.253) (0.253) (0.253)
Hispanic -0.086 0.000 -0.082 0.000
(0.053) (.) (0.053) (.)
Black — -0.643*** — -0.645***
(0.045) (0.045)
Multiracial -0.219** -0.218** -0.215** -0.223**
(0.096) (0.096) (0.096) (0.096)
HI/PI 0.152 0.158 0.155 0.151
(0.522) (0.521) (0.521) (0.521)
Asian 0.170** 0.172** 0.172** 0.172**
(0.082) (0.082) (0.082) (0.082)
SES -0.309*** -0.310*** -0.309*** -0.308***
(0.038) (0.038) (0.038) (0.038)
Constant 1.348** 1.400** 1.336** 1.406**
(0.646) (0.646) (0.645) (0.644)
School-Level Cov Y Y Y Y
Constant () 0.057** 0.057** 0.056** 0.056**
(0.026) (0.026) (0.026) (0.026)
Constant () 0.899*** 0.899*** 0.899*** 0.898***
(0.011) (0.011) (0.011) (0.011)
Observations 3435 3435 3435 3435
Standard errors in parentheses
* p < .10, ** p < .05, *** p < .01

7
Table 4. Impact of Nurturing for a Bright Tomorrow on Odds of Gifted Classification
(1) (2) (3) (4) (5)
Nurturing 1.377 1.575 2.192*** 1.910* 1.556
(0.465) (0.438) (0.628) (0.642) (0.481)
SWD 0.438 0.438 0.449 0.426
(0.263) (0.263) (0.331) (0.325)
Male 1.224 1.237 1.196 1.222
(0.265) (0.268) (0.290) (0.324)
LEP 0.865 0.858 0.923 1.127
(0.367) (0.369) (0.526) (0.676)
Am-Ind AK 1.000 1.000 1.000 1.000
(.) (.) (.) (.)
Hispanic 0.338*** 0.349*** 0.378** 0.329**
(0.135) (0.142) (0.178) (0.160)
Black 0.102*** 0.114*** 0.101*** 0.104***
(0.050) (0.056) (0.056) (0.059)
Multiracial 0.382 0.378 0.427 0.205
(0.280) (0.278) (0.321) (0.214)
HI/PI 1.000 1.000 1.000 1.000
(.) (.) (.) (.)
Asian 1.443 1.425 1.350 1.424
(0.534) (0.552) (0.646) (0.709)
SES 0.530** 0.570* 0.721 0.787
(0.152) (0.169) (0.240) (0.287)
Prior Comp Score 1.026*** 1.022**
(0.008) (0.010)
Prior FSF Score 0.998 1.008
(0.018) (0.020)
School-Level Cov N N Y Y Y
KIA N N N N Y
Constant (lnsig2u) -0.811* -1.784** -4.629 -2.607 -12.562
(0.489) (0.757) (8.392) (1.668) (31.715)
Neg2LL 838.223 762.885 752.136 596.549 505.335
Notes: Exponentiated coefficients; Standard errors in parentheses.
“Comp” is DIBELS Composite. “LNF” is DIBELS Letter Naming Fluency. “KIA” is the Kindergarten Initial
Assessment. “School-Level Cov” represents student-level covariates aggregated up to the school level.
* p < .10, ** p < .05, *** p < .01

8
Table 5. Impact of Nurturing on School-Level Gifted Counts (Incident Rate Ratios)
(1) (2) (3) (4)
Nurturing 4.149*** 5.699*** 4.528*** 2.325***
(1.043) (2.726) (2.361) (0.427)
Prior Comp Score 1.342 1.172 1.522* 0.947
(0.428) (0.344) (0.371) (0.097)
Prior FSF Score 0.877 1.184 0.397** 1.378**
(0.349) (0.554) (0.156) (0.213)
Prior LNF Score 0.982 1.027 1.284 1.043
(0.191) (0.175) (0.278) (0.111)
SWD 3.6e+10*** 1.9e+12*** 6603021.7* 45332821.9***
(1.2e+11) (1.2e+13) (53739523.8) (1.9e+08)
Male 0.119 0.030 1.251 0.002***
(0.456) (0.110) (3.991) (0.003)
LEP 1731878.6*** 786619.3*** 1.1e+08*** 100.132***
(8693398.4) (4043357.6) (4.947e+08) (159.9)
Hispanic 0.000** 0.000* 0.031 0.001***
(0.000) (0.000) (0.163) (0.002)
Black 0.001*** 0.000** 0.010 0.047***
(0.002) (0.001) (0.034) (0.043)
Multiracial 0.000*** 0.000* 0.000*** 0.000**
(0.000) (0.000) (0.000) (0.000)
HI/PI 0.000** 0.000** 0.000*** 0.000
(0.000) (0.000) (0.000) (0.000)
Asian 0.015* 0.000** 1.078 0.095
(0.032) (0.002) (2.909) (0.136)
SES 13599.1*** 4656.9*** 36827.6*** 74.7***
(46080.2) (13397.2) (116809.6) (91.3)
KIA Y Y Y Y
Observations 32 30 30 32
Notes: Model (1) includes all schools. Model (2) drops 2 IQR outlier schools. Model (3) drops two Cooks D statistic
outlier schools. Model (4) winsorizes extreme data points, such that values above the 95th percentile are replaced by
the 95th percentile and values below the 5th percentile are replaced by the 5th percentile. Standard errors appear in
parentheses.
* p < .10, ** p < .05, *** p < .01

9